Riboflavin (vitamin B.sub.2) is synthesized by all plants and many microorganisms but is not produced by higher animals. Because it is a precursor to coenzymes such as flavin adenine dinucleotide and flavin mononucleotide, that are required in the enzymatic oxidation of carbohydrates, riboflavin is essential to basic metabolism. In higher animals, insufficient riboflavin can cause loss of hair, inflammation of the skin, vision deterioration, and growth failure.
Riboflavin can be commercially produced either by a complete chemical synthesis, starting with ribose, or by fermentation with the fungi Eremothecium ashbyii or Ashbya gossypii (The Merck Index, Windholz et al., eds., Merck & Co., p. 1183, 1983). Mutants of Bacillus subtilis, selected by exposure to the purine analogs azaguanine and azaxanthine, have been reported to produce riboflavin in recoverable amounts (U.S. Pat. No. 3,900,368, Enei et al., 1975). In general, exposure to purine or riboflavin analogs selects for deregulated mutants that exhibit increased riboflavin biosynthesis, because the mutations allow the microorganism to "compete out" the analog by increased production (Matsui et al., Agric. Biol. Chem. 46:2003, 1982). A purine-requiring mutant of Saccharomyces cerevisiae that produces riboflavin has also been reported (U.S. Pat. No. 4,794,081, Kawai et al., 1988). Rabinovich et al. (Genetika 14:1696 (1978)) report that the riboflavin operon (rib operon) of B. subtilis is contained within a 7 megadalton (Md) EcoRI fragment (later referred to as a 6.3 Md fragment in Chikindas et al., Mol. Genet. Mik. Virusol. no. 2:20 (1987)). It is reported that amplification of the rib operon may have been achieved in E. coli by cloning the operon into a plasmid that conferred resistance to ampicillin and exposing bacteria containing that plasmid to increasing amounts of the antibiotic. The only evidence for rib amplification is a coincident increase in the presence of a green-fluorescing substance in the medium; the authors present a number of alternative possibilities besides an actual amplification of the operon to explain the phenomenon observed.
French Patent Application No. 2,546,907, by Stepanov et al. (published Dec. 7, 1984), discloses a method for producing riboflavin that utilizes a mutant strain of B. subtilis which has been exposed to azaguanine and roseoflavin and that is transformed with a plasmid containing a copy of the rib operon.
Morozov et al. (Mol. Genet. Mik. Virusol. no. 7:42 (1984)) describe the mapping of the B. subtilis rib operon by assaying the ability of cloned B. subtilis rib fragments to complement E. coli riboflavin auxotrophs or to marker-rescue B. subtilis riboflavin auxotrophs. Based on the known functions of the E. coli rib genes, the following model was proposed for the B. subtilis operon: ribG (encoding a deaminase)--ribO (the control element)--ribB (a synthetase)--ribF--ribA (a GTP-cyclohydrolase)--ribT/D (a reductase and an isomerase, respectively)--ribH (a synthetase).
Morozov et al. (Mol. Genet. Mik. Virusol. no. 11:11 (1984)) describe the use of plasmids containing the B. subtilis rib operon with either wild-type (ribO.sup.+) or constitutive (ribO 335) operator regions to assay their ability to complement B. subtilis riboflavin auxotrophs. From the results, a revised model of the rib operon was proposed, with ribO now located upstream of all of the structural genes, including ribG, and with the existence of an additional operator hypothesized, possibly located just upstream of ribA.
Morozov et al. (Mol. Genet. Mik. Virusol. no. 12:14 (1985)) report that the B. subtilis rib operon contains a total of three different promoters (in addition to a fourth "promoter" that is only active in E. coli). The primary promoter of the operon was reported to be located within the ribO region, with the two secondary promoters reported between the ribB and ribF genes and within the region of the ribTD and ribH genes, respectively.
Chikindas et al. (Mol. Genet. Mik. Virusol. no. 2:20 (1987)) propose a restriction enzyme map for a 6.3 Md DNA fragment that contains the rib operon of B. subtilis. Sites are indicated for the enzymes EcoRI, PstI, SalI, EcoRV, PvuII and HindIII.
Chikindas et al. (Mol. Genet. Mik. Virusol. no. 4:22 (1987) report that all of the structural genes of the B. subtilis rib operon are located on a 2.8 Md BglII-HindIII fragment and that the BglII site is located between the primary promoter of the operon and the ribosomal-binding site of its first structural gene. As described infra, Applicants show that this BglII site is actually located within the most-5' open reading frame of the rib operon, so that the 2.8 Md fragment described does not contain all of the rib structural genes. Thus, in contrast to the report of Chikindas et al., the 1.3 Md BglII fragment does not contain the ribosomal-binding site of the first structural gene; insertions at this site lead to a riboflavin-negative phenotype. Consequently, any attempt to use this BglII site to engineer the rib operon in order to increase expression, for example by replacing the 5' regulatory region with a stronger promoter, would actually destroy the integrity of the first structural gene and thus the operon as well.
Chikindas et al. (Dokl. Akad. Nauk 5 SSSR 298:997 (1988)) disclose another model of the B. subtilis rib operon, containing the primary promoter, p.sub.1, and two minor promoters, p.sub.2 and p.sub.3 : ribO(p.sub.1)-ribG-ribB-p.sub.2 -ribF-ribA-ribT-ribD-p.sub.3 -ribH. As before, it is incorrectly reported that the 1.3 Md BglII fragment contains the entire first structural gene of the operon and that this proximal BglII site maps within the primary regulatory region.